What Happens When It Will Not Read You
Every site needs a way for a refused person to start work and be paid. Most have one by accident, and it is whatever the supervisor did the first time.
The same person, one week, one reader
Three mornings out of five ended at the keypad, and each cost between six and nine minutes of somebody's time. Two people's, in fact: the person refused and the supervisor who had to be found.
A reader will refuse somebody this morning. On a site of three hundred that is a certainty rather than a risk, and the only question is what happens in the ninety seconds afterwards.
The record in “What Happens When It Will Not Read You” becomes useful only when people understand what it proves and how to correct it. For teams exploring key person dependency, the product website can connect time and project context with manager review, provided collection is proportionate, access is limited and consequential decisions remain subject to human explanation.
The second method is the arrangement that answers it. Almost every site has one and almost none designed it: it is whatever the first supervisor improvised, repeated until it became the custom, with no rule about who may use it, how often, or what gets recorded.
For an independent benchmark relevant to “What Happens When It Will Not Read You”, consult the ACCA technical resources. Use it to test notice, accessibility, security, recordkeeping, retention and exception handling against the real operating process rather than treating a device report as self-explanatory evidence.
The four arrangements in common use
A keypad code at the same terminal, entered by the person themselves. A card, carried as a backup to a biometric. A supervisor override, performed on the terminal or in the back-office software. And a paper sheet, signed at the time and typed in later by somebody in the office.
They are not equivalent. The first two keep the record in the person's own hands and keep the time accurate to the minute. The second two require another person to be present, which is why they cost six to nine minutes rather than ten seconds, and why they produce a record of what a supervisor believed rather than a record of when somebody arrived.
Why the supervisor override is the worst of the four
It is the most common and it has three separate problems. It requires a second person, who may be on the other side of the site at shift change. It produces an entry that cannot be distinguished later from a correction made for any other reason. And it puts the supervisor in the position of deciding, several times a week, whether to believe somebody about their own arrival time.
That last one does real damage over a year. It converts a technical limitation into a judgement about a person, repeatedly, in front of a queue. People stop raising it and start arriving earlier to absorb the risk, which is unpaid time nobody asked for and nobody records.
What a designed fallback looks like
It works without another person. It is available at the same terminal, in the same place, at the same moment, so the time captured is the real one.
It is decided in advance for each person who is likely to need it, rather than discovered on the morning. It records that the second method was used, which is different from recording nothing and different from recording an override. And it is used without any requirement to explain, because the reason is already known: the reader does not read this person reliably, and that was established at enrolment.
The objection, and the answer to it
The reason sites resist a self-service fallback is the obvious one: a keypad code can be shared, and the whole point of the biometric was that a credential cannot be lent.
That is true and it is a smaller problem than it looks, because the fallback is used by a known, small, named group whose usage is logged and reviewable. A code used eleven times a month by the same fitter whose enrolment record says his prints are poor is not a control weakness; it is the system working as designed. A code used by forty people with no enrolment note against any of them is a different matter, and the monitoring that distinguishes the two is a monthly report, not a policy.
What to write down
Three lines per person who needs one. Which method, why it was necessary, and who authorised it. Dated.
Then one line for the site: the circumstances in which the second method may be used by anybody — reader failure, network outage, a lost badge pending replacement — and what is recorded in each case. Without that line, every unusual morning is improvised, and the improvisations are what an auditor finds when they look at the correction history two years later.
The figure that tells you whether it is working
Count second-method entries as a share of all entries, monthly, by person and by department. Two things make it interesting.
A rising site-wide figure means enrolment quality is degrading or the environment has changed, and it is a maintenance signal rather than a discipline one. A figure concentrated in a handful of named people means the fallback is doing its job and those enrolments should be redone or those people moved to a different method permanently. Either way the number is actionable, and on most sites nobody has ever produced it.